hi, i recently deployed my livekit voice agent to ...
# ask-aid
delightful-carpenter-30895
09/13/2025, 3:10 AMhi, i recently deployed my livekit voice agent to livekit cloud; my setup is a multi agent workflow with tool calls in the process.
i’ve noticed the tool calls to write data to my mongo database doesn’t work on livekit cloud; when i run the agent locally in development, the tool calls work and i can save records to my database.
due to the limited logging available for livekit cloud agents, its a bit difficult to troubleshoot why this is the case.
are there any working examples of making external database calls in tools from livekit cloud agents? and how do i find that.
t
tall-airline-16756
09/13/2025, 3:10 AMThanks for reaching out, George. I'm an AI assistant and I'm here to help. Give me a moment while I research this topic.
tall-airline-16756
09/13/2025, 3:10 AMI'm happy to help you with using LiveKit to make external database calls through custom tools in agents! There are some great examples available for Python, including how to use SQLite with asynchronous functions to keep things running smoothly during I/O operations. For instance, the Personal Shopper Agent and Nutrition Assistant show how to create tools for tasks like retrieving order history or recording nutrition data.
For Node.js, while there aren't specific database examples provided, you can still define tools using the LiveKit SDK and connect to databases with libraries like
pgmysql2sqlite3d
delightful-carpenter-30895
09/13/2025, 3:13 AMsure, show me the example of a python agent using tools to connect to a remove mongodb or postgresql database
t
tall-airline-16756
09/13/2025, 3:14 AMI'm happy to help you with connecting LiveKit agents to external databases like MongoDB or PostgreSQL using custom tools. While there aren't direct examples for these databases just yet, the great news is that the framework supports integrating any Python library, such as
pymongo Copy code
from livekit.agents import function_tool
import pymongo
@function_tool
async def query_user_data(user_id: str):
client = pymongo.MongoClient("<mongodb://your-host:27017/>")
db = client["your_db"]
user = db.users.find_one({"id": user_id})
return str(user) or "User not found" Copy code
from livekit.agents import function_tool
import psycopg2
@function_tool
async def get_user_orders(user_id: str):
conn = psycopg2.connect(
host="your-postgres-host",
database="your_db",
user="your_user",
password="your_password"
)
cursor = conn.cursor()
cursor.execute("SELECT * FROM orders WHERE user_id = %s", (user_id,))
results = cursor.fetchall()
cursor.close()
conn.close()
return str(results)asyncpg2 Views